Accelerated Materials Design for Hydrogen Separation Membranes

field towards lower Pd contents and higher temperatures, the formation enthalpies of ... order to locate the B2 phase boundaries in the Cu-Pd-M ternar...
0 downloads 0 Views 1MB Size
Chapter 3

Accelerated Materials Design for Hydrogen Separation Membranes Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

R. Hu,*,1,2 M. C. Gao,1,2 and Ö. N. Doğan1 1National

Energy Technology Laboratory, 1450 Queen Ave. SW, Albany, OR 97321, U.S.A. 2URS Corp., P.O. Box 1959, Albany, OR 97321, U.S.A. *E-mail: [email protected]

CuPd alloys are among the most promising materials for future hydrogen separation membranes and membrane reactor applications due to their high hydrogen selectivity and permeability, improved sulfur poisoning resistance and mechanical properties. The increased permeability of certain binary CuPd compositions has been attributed to the more open bcc structure. In order to expand the ordered bcc (B2) phase field towards lower Pd contents and higher temperatures, the formation enthalpies of hypothetical B2 Cu8Pd8-xMx (x=0-8) ternary alloys were studied using first principles electronic density functional theory (DFT) at the low temperature limit. Based on the present DFT calculations, seven alloying elements were down selected including Ti, Zr, Hf, Y, La, Al and Mg for experimental verification. A total of 14 alloys were selected in order to locate the B2 phase boundaries in the Cu-Pd-M ternary system. The alloys were synthesized via arc melting, and were undergone homogenization and equlibration annealing, and then were characterized using techniques including XRD, SEM, DTA and DSC. The present approach integrates DFT calculations and experiments and allows us to identify the alloying elements that have the most potential in stabilizing the B2 phase much more rapidly compared to traditional trial-and-error experimental approach.

© 2011 American Chemical Society In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Introduction To reduce dependence on oil and emission of greenhouse gases, hydrogen is favored as an energy source for the near future (1). One way to produce hydrogen is to separate hydrogen from syngas after coal gasification (2). Membranes permeable to hydrogen only are needed to separate hydrogen from other gases. Palladium was identified as a hydrogen separation material since the mid of 1800s due to its high hydrogen permeability and selectivity (3–8). However, there are several disadvantages using palladium membrane. Hydrogen atoms occupy the interstitial sites of fcc lattice of Pd, and the interstitial solid solution undergoes a phase separation at temperatures equal to or below 293oC, forming low-hydrogen-concentration (α) and high-hydrogen-concentration (α’) solid solution fcc phases. Both phases differ from the parent random solution phase in the lattice parameters and consequently change in molar volumes can lead to internal stress and thus embrittlement (3, 9, 10). Palladium is also susceptible to contaminants such as H2S in coal derived syngas and other fossil fuels (11). Another disadvantage of Pd is its high cost. In order to develop affordable hydrogen membrane materials and to improve their H2S poisoning resistance, alloying elements including but not limited to platinum (12, 13), nickel (14–16), silver (17–23) and copper (24–29) have been extensively studied. Addition of these elements improves the surface resistance of pure Pd against poisoning in exposure to gas components, like CO, H2S and H2O (30). By alloying palladium with silver, ruthenium, copper, the phase separation of interstitial solid solution fcc phase is suppressed towards lower temperatures and the embrittlement effect can be minimized or avoided; in the mean time, hydrogen permeabilities are comparable to or greater than pure palladium for some specific compositions (31–34). Compared to the extensively studied PdAg alloys, recently PdCu alloys have attracted more attention not only because they are relatively less expensive, but also PdCu alloys exhibit better poisoning resistance to impurities (25, 26, 28, 29). High permeability values were obtained for a particular composition, Pd60Cu40 (wt%), which has the B2 structure at temperatures below 600°C (35, 36). The higher permeability is assumed due to its bcc structure that is more open than fcc structure. Bcc crystal structure has lower activation energy for hydrogen diffusion (27, 37, 38) compared to fcc structure. In order to further lower the cost of the PdCu alloys, an integrated approach that combines first-principles density functional theory (DFT) calculations and key experiments was used in this study to accelerate new CuPdM alloy design that expand the bcc phase field for hydrogen separation in syngas coal gasification.

DFT Calculations The first principles package of VASP (Vienna ab initio simulation package) (39, 40) was used to calculate the total energies using electronic DFT. Projector augmented-wave (PAW) potentials (41) were used as supplied with VASP. The Perdew-Burke-Ernzerhof (42) gradient approximation to the exchange-correlation functional was used. The Brillouin zone integrations were performed using 28 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

the Monkhorst–Pack k-point meshes (43), and a smearing parameter of 0.2 eV was chosen for the Methfessel–Paxton (44) technique. All structures are fully relaxed (both lattice parameters and atomic coordinates) until energies converge to a precision of 1 meV/at. A “high precision” setting is used. The plane-wave energy cutoff is held constant at 500 eV. The semi-core 3p, 4p and 5p electrons of transition metal elements are explicitly treated as valence. To obtain enthalpy of formation values ΔHf, a composition-weighted average of the pure elemental cohesive energies is subtracted from the cohesive energy of a given composition. The resulting energy is an “enthalpy” because its volume is relaxed at zero pressure. A 2x2x2 supercell was built and individual Pd atom was substituted by the alloying elements. Fig. 1 shows the enthalpy of formation for hypothetical B2 Cu8Pd8-xMx (x=0-8) ternary alloys for transition metals and non-transition metals. The results indicate that elements Sc, Ti, Zn, Y, Zr, Hf, La, Al and Mg are strong stabilizers under the assumptions that we ignore the competing phases in the individual ternaries and that we ignore configurational entropy effect and lattice vibration at finite temperatures. Phase stability study for the complete Cu-Pd-M ternaries at finite temperatures are necessary to draw comprehensive conclusions but are beyond the scope of the present work.

Figure 1. Enthalpy of formation of hypothetical B2 Cu8Pd8-xMx (x=0-8) alloys predicted from the present DFT calculations.

29 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Experimental Procedures Based on the present DFT calculations, we chosen 7 alloying elements (Ti, Zr, Hf, Y, La, Al and Mg) for experimental verification. Accordingly, 14 ternary alloys and two benchmark binary alloys were selected in order to locate the B2 phase field in the Cu-Pd-M ternary system on the M-poor side. The alloy compositions were listed in Table 1. The starting materials were high purity Cu, Pd, Ti, Zr, Hf, Y, La, Al and Mg elements in pellet or sponge form. The 16 alloys were prepared by melting in a vacuum arc furnace back-filled with high purity argon. Each sample weighted ~40g. After a homogenization at 900°C for 72 hours, the alloys were subjected to an equilibration annealing at 400°C for 21 days. Wavelength dispersive x-ray fluorescence spectroscopy (WDXRF) ( Rigaku, ZSX Primus II) was used for the chemical analysis. Cu-Zn alloys were used for calibrating the WDXRF. The phase identification was done using x-ray diffraction (XRD) (Rigaku, Ultima III with Jade analysis software). Optical microscopy and scanning electron microscopy/energy dispersive spectroscopy (SEM/EDS) (FEI, Inspect F50 scanning electron microscope with Oxford INCA Microanalysis) were used for microstructural characterization and microchemical analysis. The high temperature differential scanning calorimeter (DSC) apparatus (Setaram Setsys16/ 18) was used to determine the B2↔disordered fcc phase transition temperatures for these alloys. The measurements were conducted between room temperature and 1000°C with a heating and cooling rate of 10K/min in an argon atmosphere.

Experimental Results and Discussions a. Chemical Analysis The WDXRF results showed that the impurity in each sample is less than 1 wt.%, listed in Table 2. The relatively large deviation of the Mg atomic percent in one Cu-Pd-Mg sample from the nominal value might be due to the element evaporation since Mg has a lower melting point and high vapor pressure.

b. SEM/EDS and XRD Results For Cu50Pd43.75 M6.25 alloys, XRD results showed the presence of the B2 phase for M=Ti, Y, La, Al and Mg, shown in Fig. 2. All Pd27.75M6.25Cu66 alloys showed either Cu3Pd (Tetragonal, P4/mmm, L10) or Cu4Pd (Tetragonal, P42/m) as the major phase and two alloys (containing Hf and Mg) showed the B2 phase as the minor phase, listed in Table 3. The results show that the B2 phase field does not shift to the Cu-rich side in the Cu-Pd-M ternary for M=Ti, Zr, Y, La, Al, due to the presence of very stable compounds in the Cu-rich Cu-M system (e.g. Al4Cu9) or extension of stable Pd-rich Pd-M compounds into the ternary (e.g. YPd3, HfPd3, LaPd5, ZrPd3). We did not observe the binary Cu3Pd L12 phase in these samples. 30 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Table 1. Alloy compositions High Pd composition (at%)

Low Pd composition (at%)

Pd50Cu50

Pd34Cu66

Pd43.75Ti6.25Cu50

Pd27.75Ti6.25Cu66

Pd43.75Zr6.25Cu50

Pd27.75Zr6.25Cu66

Pd43.75Hf6.25Cu50

Pd27.75Hf6.25Cu66

Pd43.75Y6.25Cu50

Pd27.75Y6.25Cu66

Pd43.75La6.25Cu50

Pd27.75La6.25Cu66

Pd43.75Al6.25Cu50

Pd27.75Al6.25Cu66

Pd43.75Mg6.25Cu50

Pd27.75Mg6.25Cu66

Figure 2. XRD plots for B2-containing Cu50Pd43.75M6.25 alloys.

The volume percentage of B2 phase in each sample was estimated from the back-scattered SEM images (e.g. in Fig. 3) using ImageJ software based on the compositional contrast. Listed in Table 4 are the compositions of all phases identified using EDS and XRD as well as the estimated volume percentage of B2 phase where identified. The estimated volume percentage of B2 phase in Cu50Pd43.75 M6.25 (M=Mg, Al, and Y) is 100%, 75% and 70% at 400oC respectively. The present experiments indicate that Mg is the strongest B2 stabilizer followed by Al and Y.

31 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Table 2. WDXRF Chemical Analysis (wt.%) Ti

Mg

Zr

Y

Hf

Al

La

Alloy

Cu

Pd

Pd50Cu50

38.38

61.52

Pd34Cu66

54.82

45.08

Pd43.75Ti6.25Cu50

40.41

56.49

3.10

Pd27.75Ti6.25Cu66

58.51

38.22

3.26

Pd43.75Zr6.25Cu50

37.53

56.04

6.42

Pd27.75Zr6.25Cu66

56.50

36.34

7.15

Pd43.75Hf6.25Cu50 34.98

51.52

12.81

Pd27.75Hf6.25Cu66 52.81

33.69

13.14

Pd43.75Y6.25Cu50

37.41

56.57

5.61

Pd27.75Y6.25Cu66

55.64

37.70

6.30

Pd43.75La6.25Cu50 37.88

53.26

8.85

Pd27.75La6.25Cu66 55.80

36.93

7.23

Pd43.75Al6.25Cu50 40.94

56.26

1.81

Pd27.75Al6.25Cu66 57.89

39.07

2.11

Pd43.75Mg6.25Cu50 40.94

57.39

1.67

Pd27.75Mg6.25Cu66 57.92

40.98

0.87

Table 3. XRD results for Pd27.75M6.25Cu66 alloys Alloys

Phase 1

Pd34Cu66

Phase 2

Phase 3

Phase 4

Pd5Ti3

TiPd3

(CuPd) _FCC

Pd27.75Ti6.25Cu66

Cu4Pd_tetra

Cu3Pd_tetra

Pd27.75Zr6.25Cu66

Cu3Pd_tetra

ZrPd3

Pd27.75Hf6.25Cu66

Cu4Pd_tetra

HfPd3

Pd27.75Y6.25Cu66

Cu3Pd_tetra

YPd3

Pd27.75La6.25Cu66

Cu4Pd_tetra

LaPd5

Pd27.75Al6.25Cu66

Cu3Pd_tetra

Cu0.78Al0.22

Al4Cu9

Pd27.75Mg6.25Cu66

Cu3Pd_tetra

Cu0.7Pd0.3

PdCu_B2

PdCu_B2

Cu5.75Al4.5

32 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Figure 3. SEM back-scattered electron images of B2-containing Cu50Pd43.75 M6.25 ternary alloys (M=Ti, La, Y, Al and Mg).

33 In Modern Applications in Membrane Science and Technology; Escobar, I., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 2011.

Table 4. EDS and XRD results for Cu50Pd43.75M6.25 alloys Alloys Pd50Cu50

Downloaded by DUKE UNIV on June 19, 2012 | http://pubs.acs.org Publication Date (Web): October 31, 2011 | doi: 10.1021/bk-2011-1078.ch003

Cu50Pd43.75 Ti6.25

Cu50Pd43.75 Y6.25

Cu50Pd43.75 La6.25

Cu50Pd43.75 Al6.25

Cu50Pd43.75 Mg6.25

Cu50Pd43.75 Zr6.25

Cu50Pd43.75 Hf6.25

EDS

XRD

B2 volume %

2.45

B2

5-10%

44.43

6.63

FCC

38.82

50.52

10.65

Pd3Ti

61.45

38.55

0.00

B2

7.33

70.94

21.75

Pd3Y

20.60

61.60

17.80

FCC

65.88

34.12

0.00

B2

22.33

62.08

15.59

LaPd5

67.10

32.68

0.22

FCC

52.21

42.45

5.34

B2

75%

57.50

40.41

2.09

B2_#1

80%

46.31

45.15

8.51

B2_#2

20%

11.80

70.54

17.68

Pd0.845Zr0.155

61.05

37.49

1.47

Cu0.54Pd0.46_FCC

64.01

35.56

0.41

Cu3Pd_L12

11.63

69.71

19.20

HfPd3

Cu at%

Pd at%

51

49

60.91

36.64

48.94

M at%